See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/282889682

The construction of the finite-difference method and application

Article in AIP Conference Proceedings · March 2015

DOI: 10.1063/1.4913104

CITATION READS
1 60

2 authors, including:

Vagif Rza Ibrahimov

Baku State Univesiteti
121 PUBLICATIONS 410 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

On the construction of the effective methods for solving ODE, Volterra-integral and integro-differential equations. View project

Researchers' Gallery 😻 View project

All content following this page was uploaded by Vagif Rza Ibrahimov on 21 September 2017.

The user has requested enhancement of the downloaded file.

 The construction of the finite-difference method and application
Vagif Ibrahimov and Aliyeva Vusala

Citation: AIP Conference Proceedings 1648, 850049 (2015); doi: 10.1063/1.4913104

View online: http://dx.doi.org/10.1063/1.4913104
View Table of Contents: http://scitation.aip.org/content/aip/proceeding/aipcp/1648?ver=pdfcov
Published by the AIP Publishing

Articles you may be interested in

Applicability of the finite-difference time-domain method to photonic crystal structures
AIP Conf. Proc. 560, 99 (2001); 10.1063/1.1372719

Finite-difference methods for simulation models incorporating nonconservative forces

J. Chem. Phys. 109, 7667 (1998); 10.1063/1.477413

Application of the finite‐element method and the finite‐difference method to photothermal inspections
J. Appl. Phys. 79, 9084 (1996); 10.1063/1.362642

A general theory of linear finite‐difference equations with a few applications

J. Math. Phys. 18, 1838 (1977); 10.1063/1.523498

Finite‐Difference Methods for Partial Differential Equations

Phys. Today 14, 58 (1961); 10.1063/1.3057509

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
217.64.25.99 On: Tue, 31 Mar 2015 06:46:05
 The Construction of the Finite-difference Method and
Application
Vagif Ibrahimova and Aliyeva Vusalab
a
Doctor of science, Professor, Baku State University, Department of Computational Mathematics,
[email protected]
b
Baku State University, Department of Computational Mathematics,
[email protected]

Abstract. It is known that to find the solution to many problems of natural science, one is reduced to solving Volterra
integral equations. One of the popular methods for solving such equations is the method of quadrature. However,
methods such as quadrature methods do not have a high degree of accuracy. Thus, scientists have recently begun to
exploit different techniques that permit the construction of more accurate methods than the known methods. To this end,
we offer a number of ways to construct more accurate methods to solve Volterra integral equation. In particular, we
construct specific methods with the degree p ≤ 10 using information about the solution by considering the problem at
only one point.
Keywords: multistep methods, initial value problem, second derivative, numerical solutions.
PACS: 02.60.Lj, 02.70.Bf.

INTRODUCTION
Consider the following nonlinear Volterra integral equation of the second kind:
x
y ( x) = f ( x) + ³ K ( x, s, y ( s ))ds . (1)
x0
Integral equation (1) is commonly referred to as the Volterra integral equation of the second kind. This convention is
due to the fact that equation (1) in the linear case was thoroughly investigated by Volterra, and at a sufficiently high
level, such equations occurrence in practical problems has been studied (see, e.g., [1], [2]). Note that the singular
integral equation with a variable boundary in a particular form was studied for the first time by Abel (see, e.g., [1]).
Given that even in the linear case, the exact solution to equation (1) is not always possible, many experts have used
approximation methods to solve it (see, e.g., [3] - [7]).
The quadrature method applied to the solution of equation (1) can be written as follows (see [2]):
n
y ( xn ) = f ( xn ) + h¦ a j K ( xn , x j , y ( x j )) + Rn , (2)
j =0

where Rn is the remainder term of the quadrature method and a j ( j = 0,1,..., n) are real numbers; these numbers
are called the coefficients of the quadrature method. By discarding the remainder term, we obtain the following
method:
n
yn = f n + h¦ a j K ( xn , x j , y j ), (n = 1,2,3,...), y0 = f ( x0 ), (3)
j =0

which is called the quadrature method with a variable boundary. Although method (3) with a n ≠ 0 is implicit,
when an = 0 , this method is explicit. As a result of relations (3), each value of the kernel of the integral is
calculated n times when the number of values has the quantity n , which may increase the amount of computing
functions K ( x, s, y ) . Consequently, at each step the amount of calculation work increases, which is a major
disadvantage of quadrature methods. To eliminate this drawback of method (3), some authors have proposed the use
of methods such as the following k -step method with constant coefficients (see for example [8] - [11]):

Proceedings of the International Conference on Numerical Analysis and Applied Mathematics 2014 (ICNAAM-2014)
AIP Conf. Proc. 1648, 850049-1–850049-5; doi: 10.1063/1.4913104
© 2015 AIP Publishing LLC 978-0-7354-1287-3/$30.00

This article is copyrighted as indicated in the article. Reuse of AIP content850049-1

is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
217.64.25.99 On: Tue, 31 Mar 2015 06:46:05
 k k

¦α y
i =0
i n +i = h¦ β i f ( xn+i , y n+i ),
i =0
(4)

which has been thoroughly studied for finding the numerical solution of the following initial value problem:
y ′ = f ( x, y ), y0 = f ( x0 ). (5)
Note that the construction of methods of type (4) takes into account some equivalence with the solutions to problem
(5) and equation (1).
Here, to solve equation (1), we use the finite-difference method, which generalises many well-known methods, such
as onestep and multistep methods.
To find the numerical solution of equation (1), we intend to use partial cases of the finite-difference method with
constant coefficients:
k k

¦α y i n +i = h¦ ( β i y′n+i + βˆi yn′ +i+mi ) +

i =0 i =0
(6)
k
+h 2
¦ (γ y′′
i =0
i n +i + γˆi yn′′+i+ν i ), ( mi < 1; ν i < 1; i = 0,1, ..., k ).

In particular, for βˆi = γ i = γˆi = 0 (i = 0,1, 2, ..., k ) , from the method of (6) we obtain an ordinary k -step
method with constant coefficients and for γ i = γˆi = 0 (i = 0,1, 2, ..., k ) , we obtain hybrid methods from the
method of (6) (see, e.g., [12]- [15]). Method (6) in the general form has not been investigated. This study uses
method (6) to solve the Volterra integral equation for βˆi = γˆi = 0 (i = 0,1, 2, ..., k ) , which is commonly referred
as the second derivative finite-difference method. It is obvious that method (2.6) can be applied to solve certain
( j) ( j) (0)
problems if the values of the variables y0 , ..., yk −1 ( j = 0,1, 2; y( x ) ≡ y ( x)) are known and there is a way to
find the values yn+ k , yn′ +i + mi , yn′′+i +ν i ( j = 0,1, 2; i = 0,1, 2, ..., k ) . However, it is easy to see that an application
( j)

of method (6) to solve concrete problems is accompanied with difficulties that are not easily overcome.
Consequently, to use method (6), it is necessary to construct a special algorithm.
Consider the construction of methods of type (6) for βˆi = γˆi = 0 (i = 0,1, 2, ..., k ) . To this end, we construct a
formula for calculating the values of the function y ′(x ) and y ′′(x ) . Suppose that by some methods the solution of
the equations is found after taking into account the fact that one identity is obtained in equation (1). Then, from
identity (1) we can write:
x
y′( x) = f ′( x) + K ( x, x, y ( x)) + ³ K x′ ( x, s, y ( s ))ds, (7)
x0
x
y′′( x) = f ′′( x) + a′( x) + b( x) + ³ K x′′2 ( x, s, y ( s ))ds, (8)
x0

where a ( x ) = K ( x, x, y ( x )), b( x ) = K x′ ( x, x, y ( x )).

Finding the values of the quantities y ′n+ k and y ′n′+ k by means of formulas (7) and (8) is reduced to the calculation
of the following integral:
x

³ K (x
x0
m , s, y ( s ))ds. (9)

If quadrature formula (3) is applicable to an evaluation of integral (9), we obtain the integral sum of the variable
boundaries. For that reason, we give the prominence values y n + k −1 , y n′ + k −1 and y n′′+ k −1 , and we must find the
relationship between these values with the values of y n + k , y n′ + k , y ′n′+ k , which are unknown.
Consider the following equation:

This article is copyrighted as indicated in the article. Reuse of AIP content850049-2

is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
217.64.25.99 On: Tue, 31 Mar 2015 06:46:05
 y′′( xn +1 ) − y′′( xn ) = f ′′( xn +1 ) + a′( xn +1 ) + K x′ ( xn +1 , xn +1 , y ( xn +1 )) −
xn

− f ′′( xn ) − a′( xn ) − K x′ ( xn , xn , y ( xn )) + h ³ K x′′′3 (ξ n +1 , s, y ( s ))ds + (10)

x0
x n +1

+ ³ K ′′ ( x
xn
x2 n +1 , s, y ( s ))ds.

It is easy to show that the following can be obtained from equation (8):
xn

h ³ K x′′′3 (ξ n +1 , s, y ( s ))ds = h( y′′′(ξ n +1 ) − f ′′′(ξ n +1 )) − ha′′(ξ n +1 ) − hb′(ξ n +1 ) −

x0
(11)
ξ n +1
− hK x′′2 (ξ n +1 , ξ n +1 , y (ξ n +1 )) − h ³ K x′′′3 (ξ n +1 , s, y ( s ))ds,
xn

dK x′ ( x, s, y ( s ))
Here, the quantity b′(x) is obtained in the form b′(ξ n +1 ) = . Let us take equation (11) into
dx s = ξ n +1
account in equation (10). Then, we have:

y′′( xn +1 ) − y′′( xn ) = f ′′( xn +1 ) − f ′′( xn ) + a′( xn +1 ) − a′( xn ) +

+ b( xn +1 ) − b( xn ) + h( y′′′(ξ n +1 ) − f ′′′(ξ n +1 )) − ha′′(ξ n +1 ) − hb′(ξ n +1 ) − (12)
ξ n +1 x n +1

− hK x′′2 (ξ n +1 , ξ n +1 , y (ξ n +1 )) + ³ K ′′ ( x , s, y(s))ds + +ξ ³ K ′′ ( x
x2 n x2 n +1 , s, y ( s ))ds.
xn n +1

[ ]
Thus, the computation of the integral in the segment x0 , xn +1 using equation (8) has been reduced to calculations
in a segment [xn , xn+1 ] that calculates this integral. Moreover, this reduction was accomplished without
increasing the computational work by increasing the values of the variable n . Note that this type of scheme is
applied to the determination of the relationship between the variables y n′ +1 and y n′ .
It is known that for a sufficiently smooth function, the following holds:
k

¦α z i n+i = hz ′(ξ n +1 ). (13)

i =0
Then, by using formula (13), the interpolation polynomial of Logranzha can be employed to calculate the value
K x′′2 (ξ n+1 , ξ n+1 , y (ξ n+1 )) and an integral sum can be used to replace the integrals in equation (12). Thus, we have:
k k k k

¦α i yn′′+i = ¦α i f n′′+ i + ¦α i an′ +i + ¦α ibn +i +

i =0 i =0 i =0 i =0
k k
(14)
+ h¦¦ γ ( j)
i K x′′2 ( xn + j , xn + i , yn + i ).
j =0 i =0
After carrying out these operations, we obtain:
k k k k k

¦ α y′
i =0
i n +i = ¦ α i f n + i + ¦ α i an + i + h¦¦ γ i( j ) K x′ ( xn + j , xn + i , yn + i ).
i =0 i =0 j =0 i =0
(15)

Depending on the needs of the user, the coefficients in formulas (14) and (15) can be given different meanings. The
formulas for the calculation of y ′n′+ k can be constructed such that in methods of type (14) and (15), the terms of

This article is copyrighted as indicated in the article. Reuse of AIP content850049-3

is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
217.64.25.99 On: Tue, 31 Mar 2015 06:46:05
 type K x′ ( x, s, y ) and K x′′2 ( x, s, y ) do not participate. For this purpose, equations (7) and (8) can be rewritten as
follows:
1 k x
y′( x) = f ′( x) + a( x) + ¦ αˆi ³ K ( x − ih, s, y ( s))ds + R1 , (16)
h i = 0 x0
1 k 1 k
y′′( x) = f ′′( x) + ¦ αˆ i a( x − ih) + ¦ αˆ i K ( x − ih, x, y ( x)) +
h i =0 h i =0
x (17)
1 k
+ 2 ¦ α i ³ K ( x − ih, s, y ( s ))ds + R2 ,
h i = 0 x0
Here, αˆi , α i (i = 0,1, ..., k ) are real numbers, but R1 and R2 are the remainder terms. By replacing the integrals
in formulas (16) and (17) with an integral sum and discarding the remaining terms, we obtain methods to calculate
the values of the quantities y ′n+ k and y n′′+ k . However, in this case, the coefficients impose the following
additional conditions:
k k k

¦αˆ i = 0;
i =0
¦α i = 0;
i =0
¦ iα
i =0
i = 0.

CONCLUSIONS
As we have observed, there are different ways to solve integral equations. Recently, increasing numbers of scientists
solve these equations using multistep methods with constant coefficients. Among the most popular methods are
implicit methods. However, in the work [12], Makroglou showed that by using hybrid methods, one can construct
more accurate methods than implicit methods. It is known that second derivative multistep methods are more
accurate than the known multistep methods, and their applications to specific problems are easier than is the case
with hybrid methods. Based on this fact, the second derivative finite-difference methods in this paper are applied to
solve equation (1). In this study, we examined a method derived from formula (6) as a special case. In [15], the
application of method (6) to solving equation (1) for γ i = γˆi = 0 (i = 0,1, 2, ..., k ) was investigated, and the
defined region of stability for some methods of type (6) was also investigated.

ACKNOWLEDGMENTS
The authors express their thanks to the academician Ali Abbasov for his suggestion to investigate the computational aspects of
our problem. This work was supported by the Science Development Foundation of Azerbaijan (Grand EIF-2011-1(3)-82/27/1).

REFERENCES
1. E.M. Polishuk Vito Volterra. Leningrad, Nauka, 1977, 114p.
2. V.Volterra. Theory of functionals and integral and integro-differential equations. M., Nauka, 1982, 304p.
3. Krasnoselskiy M.A.. The approximate solution of operator equations. Ɇ.: Nauka, 1969, 210 p.
4. Y.D. Mamedov, Ashirov S.A. Methods of successive approximations for the solution of operator equations. Ashgabat, 1980,
120 p.
5. A.F. Verlan, V.S. Sizikov. Integral equations: methods, algorithms, programs. Kiev, Naukovo Dumka, 1986, 384 p.
6. Manzhirov A.V. Polyanin A.D. Handbook of Integral Equations: Methods of solutions. Moscow: Publishing House of the
"Factorial Press", 2000, 384 p.
7. Atkinson K.E. A survey of numerical methods for solving nonlinear integral equations. J. of integral equations and
applications, 1992, No 1, v.4, p.15-46.
8. Ch. Lubich. Runqe-Kutta theory for Volterra and Abel Inteqral Equations of the Second Kind. Mathematics of computation,
volume 41, number 163, July 1983, p. 87-102.
9. Brunner H. The solution of Volterra integral equation of the first kind by piecewise polynomials. J.Math . and Appl., 1973,
No12, p.295-302.
10. G. Mehdiyeva, M.Imanova On an application of the finite-difference method. Bulletin of the University, 2008, ʋ2, p.73-78.

This article is copyrighted as indicated in the article. Reuse of AIP content850049-4

is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
217.64.25.99 On: Tue, 31 Mar 2015 06:46:05
 11. Mehdiyeva G.Yu., Imanova M.N., Ibrahimov V.R. An application of the finite-difference method. Bulletin of the University,
2009, ʋ3, pp.101-108.
12. Makroglou A. Hybrid methods in the numerical solution of Volterra integro-differential equations. Journal of Numerical
Analysis 2, 1982, pp.21-35.
13. H. Brunner. Imlicit Runge-Kutta Methods of Optimal oreder for Volterra integro-differential equation. Methematics of
computation, Volume 42, Number 165, January 1984, pp. 95-109.
14. Mehdiyeva G.Yu., Imanova M.N., Ibrahimov V.R. On one generalization of hybrid methods // Proceedings of the 4th
international conference on approximation methods and numerical modeling in environment and natural resources Saidia,
Morocco, may 23-26, 2011, P. 543-547.
15. G.Yu. Mehdiyeva, V.R. Ibrahimov, M.N. Imanova On the construction test equation and its applying to solving Volterra
integral equation. Mathematical methods for information science and economics, proceedings of the 17th WSEAS
International conference an Applied math. (AMATH 12), pp. 109-114.

This article is copyrighted as indicated in the article. Reuse of AIP content850049-5

is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
217.64.25.99 On: Tue, 31 Mar 2015 06:46:05
View publication stats